Why would twenty questions on the same state cost about as much as one?

Because the state is read once. The claim is that Jev scores all of the options you defined against a single pass over the input, so extra questions over the same state add options rather than another read. That is the pricing model, not a benchmark result.

Is cheap input with free output the real deal or just list pricing?

It is vendor list pricing restated in the description, and the entry flags it as exactly that. Free output makes sense for a model emitting option keys and probabilities rather than prose, but nothing here verifies the rate or how an input token is counted.

Does one pass still hold up when every question has many options?

Unaddressed. The capture has no transcript and no table, so the case that worries you — many questions multiplied by long option lists, where the bill scales with options times state — is not something this entry can speak to. Check the vendor's own figures.

How is this different from a small chat model constrained to a fixed enum?

The description sells typed options plus a probability per option, with no output tokens to parse and no schema errors to handle. Whether that beats a small model pinned to an enum is a comparison this capture does not make.

Who made the video, and how does it describe Jev's output?

ZAUEY (Claire Zau) published it, describing Jev as picking from answer options you define and returning a probability for each.

How is input priced in the description?

$42 per billion input tokens, with output priced at zero. The entry records that as the description restating vendor pricing.

What disclaimer does the description carry?

That it is not investment advice, noted alongside the price and the twenty-questions claim.

CommunityJev Media & Talks

Jev: System One model

ZAUEY on a model that picks from options you define and returns a probability, and why many questions on one state stay cheap.

See highlights

Open youtube.com

Format: Video
Published by: Community
Author: ZAUEY (Claire Zau)
Last verified: Sep 2026

Tagscommunity video explainer cost

Works forJev for Classification

Used inJev for Software & DevTools

Highlights

Describes Jev as picking from answer options you define and returning a probability for each.
The description says 20 questions cost about the same as one, because the state is read in a single pass.
It prices input at $42 per billion tokens and output at zero.

Watch out

The price and the 20-question claim are the video description restating vendor pricing. Not investment advice, as the description notes.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

YouTube: ZAUEY on a model that cannot write, and why twenty questions cost about the same as one.

Hacker News: Simon Willison's overview of Noul, Choice, and Score, and where a decision model fits.

benwerd53 pointsJev introduces a new shape of LLMRead the thread on Hacker News

Hacker News: swyx posts the Latent Space interview with Diogo Almeida.

swyx4 pointsJev: System One Models for Prod, Not God – With Diogo Almeida, CEO, TypeSafe AIRead the thread on Hacker News

X: Diogo Almeida in five minutes

Related terms

Glossary definitions related to this entry.

More Jev resources like this

IBM Developer: probabilities, not words

Max Jesch, on IBM Developer, walks Jev as a System One model that returns probabilities for classification instead of generating text.